This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch/rfc] Moving constraint definitions to the machine description
On Thu, 12 Jan 2006, Zack Weinberg wrote:
> I think that only C identifier characters should be allowed in
> MI constraint names (and point to the above interface as a practical
> reason why); you think punctuation should be allowed too. I think that
> there is no need to enforce the convention that G-P as a leading letter
> in a constraint name are reserved for CONST_DOUBLEs or CONST_INTs (and
> point to DJ's m32c where it would be appropriate to relax that convention);
> you want extra hoops to be jumped through if the convention is to be
> violated. I do not wish to argue either of these points with you further;
> you're not going to change my mind, nor I yours. Instead I ask that
> other interested parties weigh in.
FWIW, restricting the target-adjustable constraints to
alphanumeric (perhaps adding _) sounds reasonable to me and I
don't really think relaxing or expanding the existing naming is
important.
I don't really like the tweaks of unwritten facts that I think I
saw suggested somewhere in the thread, like making e.g. m or r
be target-adjustable.
brgds, H-P